roadmap_ref.pdf — Extracted Text & Logic
Source: 6-page PDF, no text layer — each page is a photo of a handwritten notebook page (pencil) or office whiteboard (blue marker). Pages 1, 4, 5, 6 are paper; pages 2 and 3 are whiteboards. All pages were photographed rotated 90°.
[?] marks a word I could not read with confidence.
Legend (inferred from repeated use across pages)
| Token | Reading | Evidence |
|---|---|---|
WF | Workflow (automation) | "End to End Automation with WF" (p5), "E2E with WF" (p5) |
E2E | End to End | p5 |
A, A1…A6 | Agent — each spoke off an OS is an agent | p1 labels every Mktg OS spoke "A"; p2 numbers them A1…A6 |
MVP | Minimum viable scope, phase 1 | p3 "③ Happy Case (MVP)"; p6 "MVP = YT/Meta/FB/Insta/WA" |
ES | End State | p3 "④ End State (V1)"; p6 uses ES for everything deferred past MVP |
I / E | Phase tag on a deliverable — Immediate vs End-state | p5 "STATIC — I / GIF — I / Video — E" |
LFC | Long-form content | p4 "LFC → Scan + Planning + Blog + Execution" |
CLCM | Customer lifecycle management | p2 "Onboarding CLCM" |
NBA | Next Best Action | p2 |
KKH [?] | Planning framework — "Where & What" | p3 |
Page 1 — Marketing OS (the hub diagram)
Title, top-left: Creative Creation
Centre node: Mktg OS. Five spokes, each annotated A (= Agent):
| Spoke | Notes |
|---|---|
| ATL OS | — |
| BTL OS | ← Process Automation |
| Analytics OS | — |
| Program OS | expanded on page 2 |
| Digital OS | expanded on page 3 |
Digital fans out to:
- Branding →
h + M[?] - Performance
- Organic (web / app)
- GMB – 3050 (3050 = listing count)
- Personalised Experience
- FUNNEL – CRO → Custom Exp[erience]
- Campaign mgmt → Retention
- Referral →
Data layer (bottom of page), feeding the OS:
Outside world DATAIIFL DATA→ (CRM)MANUAL (FLEQ)[?]
Logic: one Marketing OS, five sub-OSes, each sub-OS driven by agents; two data sources (owned IIFL/CRM + external) plus a manual feed.
Page 2 — Program OS: Retention & Acquisition (whiteboard)
Centre: Programs → splits into Retention and ACQ (circled).
Retention (70%)
Routes to → IIFL BASE and → Non-IIFL BASE → Lead (Not conv.)
A1 — Active base
| ID | Play | Target |
|---|---|---|
| A1.1 | Add-on Loan | 500 cr |
| A1.2 | Top up | 2.5k |
| A1.3 | Cross Sell | ₹–? |
| A1.4 | Refer | 1000 |
| — | Renew | 500–700 cr [?] |
| — | Interest payment | 3000 / 1 cr [?] |
Delivery channels for A1: WA Bot / App Adoption / Digital Web
A2 — Inactive base (bucketed by recency)
- Winback — 0–6 m
- Closure — 6 m – 24 m
- Rechurn — > 24 m
- Competition Base — Surgical Strike
- Referral
The agent chain (A3 → A6)
⇒ Segments (A3) ─┬→ Micro-segmentation — ✗ ┐
├→ Real-time Engagement — ✗ ├ CRM
└→ Intelligence — ✗ ┘
A4 → Communication Creation → AI Avatar | ATL
A5 → Rollout → SMS / WA / RCS / App push / In-App | Clevertap, Netcore, Humanoid
A6 → Analytics
→ Richness Data (P) ⟨ Lead / Conv / Post [?] ⟩
↳ Onboarding CLCM
The ✗ marks are gaps — capabilities that do not exist today.
ACQ (Acquisition)
- Referral (Easy Rewards)
IIFL BASE → LEAD→ ✗ Loyalty- ↳ Allocate to Branch
- ↳ Branch to Connect
- ↳
C – NC→ ✗
- DSA — Physical world
- Share & Win [?] — Digital world ⎫
- Infl. (influencer) ⎭ → NON IIFL BASE
Target-state capabilities (right column)
⇒ Data Enrichment · Onboarding CLCM · Personalization · Real-time Engagement · Intelligence Agent · NBA
Page 3 — Digital OS (whiteboard)
Banner: NO AGENCY — "Built with Co-Pilot – LLM".
The thesis of the whole board: replace agency work with LLM-driven workflows.
Planning column (left)
- Deep Analytics / Call Analytics
- KKH [?] = Planning — Where & What
- Budgeting · Execution
- Discovery – Th. · Creation · Budget – Dynamic · Setup · Rollout · CRO · Monitoring · Optimization
- Assumption → Reality
- Platform Discovery
- Competition Research
Buying stack (same three vendors used twice)
Relid [?] / Zocket / Pixis — circled DWAO
| Branding | Performance |
|---|---|
| DV360-G | Google — UAC / Search / Display |
| Meta | Meta → 7 streams |
| Publishers | Affiliate |
| OTT | OEM |
| Infl. → ASP | SST |
| Audio | |
| ASP |
Spokes off Digital
- ORM → Meltwater, LocoBuzz
- Digital OS → DB / Integration
- Research / Feedback → Customer / Emp.
- Digital Partnerships (multiple) → Discovery / Outreach / Hyperlocal
- CRO → Full-funnel mgmt → Customer / Employee — via Clevertap / Netcore / Rubic [?] / Humanoid
- Hyperlocal (Web / App) — Single Interface
- Organic SEO (Web / App) — Off-page / Draft / Syndication / Media / YT
- → Personalised Experience
- → Scale Growth (circled):
- Th. Discovery – KKH → Nat / State / City / Micromarket
- Page Structure
- Personalised exp.
- Content — LFC / Image / Video / Voice / Vernac.
- Publish / Rollout
- CRO · Monitoring · Optimization
- Categorised by: Product · Genre · User · RTB · Brand Ambassador [?]
- ATL (Creative) · Social · Digital Distribution · PR / Grapevine [?] · Customer Profiling
Build sequence (right margin — applies to every workstream)
- Repetitive / Volume [?]
- Expected outcome
- Happy Case (MVP)
- End State (V1)
Page 4 — Scope by channel + creative architecture
1) ATL → Creative → 15000 → Static | GIF | Video ... f/HO
→ Social → Creative + Analytics + Planning + Scan – Planning
Trend spotting — Paid + Infl. spotting
→ ORM → own
→ LFC → Scan + Planning + Blog + Execution
2) Architecture
Creative → Hivocal [?] | Zocket | Righton — Partner ecosystem
Adobe Firefly | Canva |
① Creative (own)
-
→ Template + UGC → under guidelines
-
→ Diff Approaches → Diff Themes
-
→ Localization + Topicality
-
→ FACTORY
↳ Discovery → Content → RBI → Creation → RBI Approval ← Vernac ↑ RBI -
Req. Capturing
RBIappears three times as a compliance checkpoint in the creative factory loop — content, creation and approval each pass an RBI gate before the vernacular versions go out.
Page 5 — The build backlog (two sheets)
Sheet A — programme list
- Creation End to End with WF — LFC
- Social Media End to End Automation with WF
- ORM — End to End Automation with WF
- Marketing OS — Architecture
- (sketch: boxes labelled
Repeat,C→H,?,VA— 8 R [?])
Sheet B — numbered deliverables (the actual roadmap)
| # | Deliverable | Phase tag |
|---|---|---|
| ① | Creative (BTL + Perf + Social + Organic) — Static | I |
| — GIF | I | |
| — Video | E | |
| E2E with WF | ||
| ② | Blog — E2E with WF | I + E |
| ③ | SM (Social Media) — E2E — A with WF | I + E |
| ④ | Marketing OS — Architecture | |
| ⑤ | Dashboard / Analytics | I + C |
| ⑥ | ORM | |
| ⑦ | Creative Repository |
Page 6 — Per-workstream spec: MVP vs End State
Creative → WF Automation — Brand Guidance
Blog
- Tonality
- Imagery / Aspect Ratio
- Guard Rails
- Size – Weight
- Vernac
- Template — Standard
- Template — Dynamic
ES— Content – Persona – Dynamic
NO AGENCY
SM (Social Media)
MVP = YT | Meta | FB | Insta | WA
ES = ✗ | LinkedIn
ES = Infl. + Avatar
ES = Agency Automation
ES = ORM
NO AGENCY
Feedback loops (right column, all tagged ES)
- Learn from what we have done → ES
- Scan competition to get insights → ES
- Scan category to get insights → ES
The logic, distilled
- One
Mktg OSsits at the centre, decomposed into five sub-OSes — ATL, BTL, Analytics, Program, Digital. Each sub-OS is driven by agents (A1…A6), not by an agency.NO AGENCYis written twice on p3 and twice on p6 — it is the organising principle, not an aside. - Two data feeds into the OS: owned
IIFL DATA(CRM) andOutside world DATA, plus a manual feed to be eliminated. - Program OS = lifecycle. Retention (70%) is split Active (A1) vs Inactive (A2) with recency buckets; acquisition (ACQ) runs referral, DSA, digital share-and-win and influencer, all converging on the Non-IIFL base. A3–A6 are the shared pipeline: segment → create → roll out → measure.
- Digital OS = channel engine. A single planning frame (KKH: where & what → budget → execute) feeds a fixed vendor stack (Zocket / Pixis / DWAO) across branding and performance, with SEO / hyperlocal "Scale Growth" drilling Nat → State → City → Micromarket.
- Every workstream ships through the same four gates:
repetitive volume → expected outcome → happy case (MVP) → end state (V1).
The
I/E/ES/MVPtags on pages 5 and 6 are that gate applied deliverable by deliverable. - Creative is a factory, not a brief. Discovery → Content → RBI → Creation → RBI → Approval, with vernacular fan-out and templates (standard + dynamic) under brand guard rails.
- The gaps are marked explicitly with
✗: micro-segmentation, real-time engagement, intelligence, loyalty, C–NC, LinkedIn. Those are the build list.
Rendering the pages yourself
The PDF has no text layer. To reproduce the upright images (requires PyMuPDF):
python -c "import fitz; d=fitz.open('roadmap/roadmap_ref.pdf'); [d[i].get_pixmap(matrix=fitz.Matrix(3,3).prerotate(r)).save('page%d.png'%(i+1)) for i,r in enumerate([270,270,270,270,90,270])]"
Pages 1–4 and 6 need prerotate(270); page 5 needs prerotate(90) for its
left sheet and 270 for its right sheet (the two sheets face opposite ways).